0 Active Nodes when running Hadoop and Yarn in pseudo-distributed mode on CentOS 7

1
votes

Symptoms:

  • sbin/start-dfs.sh and sbin/start-yarn.sh execute without errors, but localhost:8088 shows 0 active nodes.
  • mapreduce jobs fail to connect to ResourceManager at /0.0.0.0:8032
  • Password-less (public key) ssh is enabled and working.
  • core-site.xml, hdfs-site.xml, mapred-site.xml, and yarn-site.xml are all configured correctly.
  • OS is CentOS 7 running on Oracle VirtualBox.
1
hadoopcentos7

1 Answers

0
votes

Solution:

  • Not stated in the tutorial I was following: CentOS needs to be configured to allow a connection to the local machine.
  • I edited the file /etc/hosts with root permissions.
  • I added a line containing my machine's numeric ip address, a space, and then the machine's host name (as configured during the CentOS installation).
  • (In my virtual configuration, this was the ip address of the virtual machine's "host-only adapter".)
  • I stopped, and then restarted, dfs and yarn.
  • I checked localhost:8088: 1 active node.
  • My mapreduce jobs now executed successfully.